· web viewthe relationship between depression and biologic treatment response in rheumatoid...

Depression and biologic treatment response

The relationship between depression and biologic treatment response in rheumatoid arthritis: An analysis of the British Society for Rheumatology Biologics Register. Faith Matcham, Rebecca Davies2,3, Matthew Hotopf1,4, Kimme Hyrich 2,3, Sam Norton5,6, Sophia Steer6, James Galloway6.

1. King’s College London, Department of Psychological Medicine, Institute of Psychiatry, Psychology and Neuroscience, London, UK.

2. Arthritis Research UK Centre for Epidemiology, Centre for Musculoskeletal Research, Institute of Inflammation and Repair, the University of Manchester, UK.

3. NIHR Manchester Biomedical Research Centre, Central Manchester Hospitals NHS Foundation Trust, Manchester Academic Health Science Centre, UK.

4. South London and Maudsley NHS Foundation Trust, Maudsley Hospital, London, UK.

5. King’s College London, Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience, London, UK.

6. King’s College London, Department of Academic Rheumatology, London, UK.

Correspondence address:

Dr Faith Matcham,Institute of Psychiatry, Psychology & Neuroscience, King’s College London,Weston Education Centre,10 Cutcombe road,London SE5 9RJ [email protected](+44) 2078480868

1

mailto:[email protected]


ABSTRACT

ObjectiveTo investigate the relationship between depressive symptoms and treatment

response and disease activity over a one-year follow-up.

MethodsData from the British Society for Rheumatology Biologics Register were used,

representing 18,421 RA patients receiving biologic treatment. Depressive symptoms

were identified through one of three assessments: reporting a history of depression;

the Medical Outcomes Survey 36-item Short Form (SF36); or the EuroQol (EQ5D).

Logistic regression analyses examined the relationship between baseline depressive

symptoms and odds of good treatment response by 1-year. Multilevel models

addressed the association between baseline depressive symptoms and disease

activity outcomes over 1-year follow-up, adjusting for age, gender, disease duration,

comorbidities, and baseline disease activity and physical disability.

ResultsDepression symptoms at biologic treatment initiation were associated with 20-40%

reduced odds of achieving a good treatment response at 1-year. Depressive

symptoms at baseline also associated with reduced improvement in disease activity

over the course of follow-up. Patients with a history of depression or reporting

symptoms of depression according to the EQ5D showed reduced improvement in

tender and swollen joints, patient global assessment (PGA) and erythrocyte

sedimentation rate (ESR) over 1-year follow-up. Patients with depression symptoms

according to the SF36 showed reduced improvement in tender and swollen joints,

but not ESR or PGA.

ConclusionExperiencing symptoms of depression at the start of biologics treatment may reduce

the odds of achieving a good treatment response, and reduce improvement in

disease activity over time. Depression should be managed as part of routine clinical

care to optimise treatment outcomes.

Keywords: Rheumatoid Arthritis, Biological therapies, Depression, Epidemiology,

Quality of life, Mental health services, Statistics

2


INTRODUCTIONDepression is prevalent in RA, with meta-analysis evidence suggesting a 17% point

prevalence according to diagnostic interview [1]. Recent evidence has highlighted

depression symptoms as a prognostic psychomarker for poor rheumatological

outcomes, with symptoms of depression and anxiety associated with worsened

disease activity, physical function, and reduced response to Disease Modifying Anti-

Rheumatic Drug (DMARD) and glucocorticoid treatments [2,3]. Depression is rarely

measured in rheumatological research [4]; assessment of mental health is usually

limited to assessment of mental health domains on health-related quality-of-life

(HRQoL) questionnaires such as the Medical Outcomes Survey 36-item Short-Form

(SF36) [5] or the EQ5D [6].

The development of biologic treatments for RA have revolutionised the management

of RA; in comparison to conventional DMARDs, biologics contribute to increased

likelihood of disease remission, and significantly improve physical function [7,8].

There is, however, a dearth of research examining the relationship between

depression and long-term disease outcomes in RA [3]. The impact of comorbid

depression on biologic treatment response has previously been investigated in a

United States register; the authors reported an association between depressive

symptomatology and likelihood of remission at 6-month follow-up, purported to be

driven by an association between depression and subjective experiences of disease

[9]. Their assessment was limited to a one-item depression comorbidity tick-box,

which may result in high false-positive rates due to poor specificity [10]. Analysis of

the Norwegian Disease-Modifying Anti-Rheumatic Drug (NOR-DMARD) database

identified an association between depressive symptoms at the start of treatment with

biological or synthetic DMARDs, and reduced risk of remission and higher pain and

global assessment at 3- and 6- month follow-up [11]. To date, there is stronger

evidence for a relationship between depressive symptoms and subjective

experiences of disease such a patient global assessment (PGA) and pain [2,11–13],

contributing to a growing focus on re-defining remission in RA [14].

The present study seeks to examine the longitudinal association between depressive

symptoms and treatment response in a UK national register of RA patients starting

their first biologic. In comparison to previous research [9], this study utilises three

measures of depression symptoms: history of depression comorbidity tick-box, the

3


mental health domain of the SF36, and the depression/anxiety item of the EQ5D,

providing an opportunity to investigate the differential relationships between

depression symptoms and treatment outcomes based on mental health assessment

strategy. The aims are: 1) to examine the relationship between baseline depression

symptoms and biologic treatment response over 1-year follow-up; 2) to evaluate the

relationship between baseline depression symptoms and disease activity over 1-

year.

METHODS

ParticipantsThis study presents an analysis of the British Society for Rheumatology Biologics

Register for Rheumatoid Arthritis (BSRBR-RA; [15]). The BSRBR-RA is a national

prospective register of RA patients starting a new biologic, and contains data from

18,421 patients enrolled since its inception in 2001. To be eligible for inclusion in the

BSRBR-RA, patients must meet UK guidelines for commencing a biologics:

sustained active RA (defined as scoring >5.1 on the DAS28 at two timepoints a

month apart); and failure to respond to ≥ conventional Disease modifying Anti-

Rheumatic Drugs (DMARDs) including methotrexate over a ≥6 month timeframe

[16]. A range of clinical, demographic and psychological assessments are taken at

baseline, 6-monthly intervals for the first 3-years of follow-up, and yearly thereafter.

We limited our analysis to first biologic exposure only.

Assessments

Depression Symptoms

Three measures of depressive symptoms are available in the BSRBR-RA database:

upon enrolment, all patients are asked if they have ever had or received treatment

for depression, which was used as an indicator of history of depression. The SF36

[5] was used as an assessment of HRQoL between 2001-2008, in the first 11,937

enrolled patients. A threshold of ≤40 on the normed mental health (nMH) subscale of

the SF36 has been shown to have a 92.6% sensitivity and 73.2% specificity for

identifying depression in patients with RA [17]. Responses to the SF36 were

categorised using this threshold, to represent patients with low HRQoL. The EQ5D

[6] was introduced to the BSRBR-RA in 2005 and became the only HRQoL

assessment from 2010 onwards. The EQ5D has one item specific to mental health,

4


allowing patients to identify whether they are feeling “not depressed/anxious”,

“moderately depressed/anxious”, or “extremely depressed/anxious” today. Evidence

suggests that one-item mood screeners have 84% sensitivity and 65% specificity

[18], and this item has been previously used to predict longitudinal DAS28 and HAQ

outcomes, and prednisolone treatment response in RA patients [2].

Treatment Response

The primary outcome of interest was 1-year treatment response, measured by the

European League Against Rheumatism (EULAR) guidelines [19]. Based on their 1-

year EULAR response, patients were categorised into those demonstrating a good

treatment response and those with a suboptimal treatment response

(none/moderate response).

Disease activity

Secondary outcomes were disease activity (measured via the DAS28), and its

composite parts: tender joint count (TJC); swollen joint count (SJC); patient global

assessment (PGA); and erythrocyte sedimentation rate (ESR); all measured at 1-year

follow-up. TJC and SJC underwent square root transformation and ESR data were

log transformed for analysis.

Statistical AnalysisAlthough data were available for three years of follow-up, only data until the first

year of follow-up were included. This ensured a focus on first biologic exposure,

eliminating bias introduced by patients switching biologics due to a lack of treatment

response. All analyses were conducted on Stata v14. Clinical and demographic

characteristics of patients having a good treatment response by 1-year were

compared with those with no/moderate treatment response using means and

standard deviations, with statistically significant imbalance determined using t-tests

for continuous variables and Chi- squared tests for categorical data.

Aim 1: The impact of baseline depressive symptoms on 1-year biologic treatment

response

Logistic regression models were performed in 2 stages: unadjusted (model 1), then

adjusted for age, gender, disease duration, baseline DAS28, baseline HAQ, and

number of comorbidities (model 2). Logistic regression estimates the odds of a

binary outcome (i.e. having a good treatment response), based on several predictor

5


variables (i.e. baseline depression). In all models, baseline depressive

symptomatology was entered as the predictor variable: history of depression

(yes/no); SF36 nMH subscale (≤40/>40); EQ5D (no/moderate/extreme)). Treatment

response at 1-year (none/moderate vs. good) was the outcome variable in primary

analyses. Odds ratios (OR), p-values and 95% confidence intervals estimated

whether the presence of depressive symptoms at biologic initiation was associated

with increased odds of having a good treatment response at 1-year. Multiple

imputation was used to address baseline missing data for DAS28, disease duration

and HAQ.

Aim 2: Relationship between baseline depressive symptoms and time-course

disease activity over 1-year

The relationships between baseline depression and 1-year DAS28, TJC, SJC, PGA

and ESR were examined using multilevel longitudinal models, pooling data across

the timepoints (baseline, 6-months and 1-year) [20]. Multilevel modelling handles

hierarchically nested data, accounting for missing data, and both between- and

within- participant variation over time, and multiple imputation was used to address

baseline missing data for DAS28, disease duration and HAQ.

Output from multilevel models includes unstandardized maximum likelihood

estimates (B coefficients), which estimate the magnitude and direction of change in

an outcome variable according to a reference group (no depressive symptoms at

baseline). In addition to depressive symptoms as a predictor variable and DAS28

(and its composite parts) as outcome variables, multilevel models included time as

a continuous variable coded as 0 at baseline, 1 at 6-months and 2 at 12-months,

and the interaction between time and baseline depressive symptoms, to examine

whether change over time is different between people with and without symptoms of

depression at baseline. A random intercept and random time slope allowed for

variation in the baseline level of the outcome and the rate of change in the outcome

between individuals. The random effects were allowed to correlate, which means

that some control for the baseline level of the outcome is included in the model

even though it is not included as a covariate – e.g. a positive correlation would allow

for increasing variability in the outcome variable over time [21]. Multilevel models

were created in two stages: 1) including only baseline depression symptoms, time

and an interaction between time and depression symptoms, plus random effects;

6


and 2) additionally adjusting for age, gender, disease duration, comorbidities, and

baseline physical activity (measured via the Health Assessment Questionnaire

(HAQ) [22]). Covariates were selected based on theoretical relevance.

RESULTS

Missing data

Figure 1 shows the data available for analysis for the primary outcome (treatment

response) and secondary outcome (disease activity), in relation to the different

methods of depressive symptom measurement. Missing response status/disease

activity outcome data at 1-year was associated with increased BMI, and lower

baseline DAS28, TJC, SJC, ESR and HAQ, shown in supplementary table t1.

Baseline depression symptoms, according to all three measurements available, was

not associated with missing outcome data.

Participant characteristics

Data from 18,421 patients enrolled in the BSRBR-RA by December 2015 were

included in this analysis (figure 1). Table 1 displays baseline demographic, clinical

and psychological variables for all patients. The mean age was 56.4 years, 76.4%

were female, with a mean disease duration of 12.6 years and mean DAS28 of 6.4.

By 6-months, 3,638 patients had achieved a good treatment response. At 1-year, a

total of 5,271 (34.3%) of patients having reached a DAS28 of ≤3.2 and an

improvement in DAS28 from baseline of >1.2.

At 1-year follow-up, 17.9% of patients were identified as switching biologic. Biologic

switching was significantly higher in patients reporting a history of depression, and

depressive symptoms according to the SF36 and EQ5D at baseline (supplementary

table t3).

7


Figure 1. Summary of the number of patients available for each outcome analysis. ^percentages calculated from the total sample.

8


Table 1. BSRBR-RA cohort baseline characteristics for total group, and according to baseline depressive symptoms.

History of depression SF36 nMH depression symptoms EQ5D depression symptomsTotal Sample

(N=18,421)No

(N=14,426; 73.7%)

Yes (N=3,669;

20.3%)p-value

No (N=5,388 (45.1%))

Yes (N=6,549,

54.9%)p-value

None (N=2,761 (44.6%))

Moderate (N=2,975 (48.1%))

Extreme (N=453

(N=7.3%)) p-valueAge, M (SD) 56.4 (12.4) 56.6 (12.6) 55.2 (11.4) <0.0001 55.9 (12.1) 56.1 (12.3) 0.07 57.6 (12.9) 57.1 (12.4) 54.8 (11.5) <0.0001Female Gender, N (%) 14,065 (76.4) 10,799 (74.9) 2,999 (81.7) <0.0001

4,056 (75.3) 5,057 (77.2) 0.01 2,094 (75.8) 2,299 (77.3) 354 (78.2) 0.329BMI, M(SD) 27.3 (6.6) 27.1 (6.5) 28.1 (6.8) <0.0001 26.8 (6.4) 27.0 (6.2) <0.0001 27.7 (7.1) 28.2 (7.3) 29.6 (7.5) <0.0001Disease Duration, years, M(SD) 12.6 (9.7) 12.7 (9.8) 12.3 (9.6) 0.001

13.1 (9.5) 12.9 (9.7) 0.007 12.1 (10.0) 11.8 (10.0) 9.8 (9.6) <0.0001Smoking, N(%)Current 3,733 (20.3) 2,762 920.2) 922 (27.3) <0.0001 1,053 (19.7) 1,546 (23.8) <0.0001 420 (17.9) 593 (23.4) 118 (30.2) <0.0001Ex-smoker 6,584 (35.7) 5,176 (37.9) 1,312 (38.9) 2,119 (39.6) 2,415 (37.1) 897 (38.2) 979 (38.6) 135 (34.5)Never smoked 6,990 (38.0) 5,736 (42.0) 1,141 (33.8) 2,176 (40.7) 2,546 (39.1) 1,031 (43.9) 967 (38.1) 138 (35.3)Comorbidity, N(%)* 9,988 (55.5) 7,565 (53.3) 2,281 (63.5) <0.0001 2,747 (52.0) 3,561 (55.4) <0.0001N comorbidities, M(SD)* 0.9 (1.0) 0.8 (1.0) 1.1 (1.2) <0.0001

0.8 (1.0) 0.9 (1.0) <0.0001 0.9 (1.1) 1.0 (1.1) 1.1 (1.3) <0.0001Treatment Type, N(%)Etanercept 5,356 (29.1) 4,232 (29.3) 1,039 (28.3) <0.0001 1,797 (33.5) 2,275 (34.7) 0.001 298 (10.8) 259 (8.7) 39 (8.6) <0.0001

Infliximab 4,249 (23.1) 3,348 (23.2) 853 (23.3) 1,587 (29.5) 2,062 (31.5) 72 (2.6) 122 (4.1) 19 (4.2)Adalimumab 5,024 (27.3) 4,044 (28.0) 904 (24.6) 1,992 (37.0) 2,196 (33.5) 840 (30.4) 993 (33.4) 162 (35.8)Rituximab 1,650 (9.0) 1,208 (8.4) 393 (10.7) 12 (0.22) 16 (0.24) 654 (23.7) 813 (27.3) 100 (22.1)Tocilizumab 1,008 (5.5) 709 (4.9) 267 (7.3) - - 411 (14.9) 378 (12.7) 68 (15.0)Certolizumab 1,115 (6.1) 870 (6.0) 210 (5.7) - - 478 (17.3) 404 (13.6) 65 (14.4)Infliximab Biosimilar 19 (0.1) 15 (0.1) 3 (0.1)

- - 8 (0.3) 6 (0.2) 0 (0.0)Baseline disease statusDAS28, M(SD) 6.4 (1.1) 6.4 (1.1) 6.4 (1.1) <0.0001 6.5 (1.0) 6.6 (1.0) <0.0001 5.9 (1.2) 6.1 (1.1) 6.3 (1.0) <0.0001

TJC, M(SD) 15.2 (7.5) 15.0 (7.5) 15.9 (7.5) <0.0001 15.2 (7.3) 15.9 (7.4) <0.0001 13.6 (7.5) 15.0 (7.5) 16.2 (7.3) <0.0001

9


SJC, M(SD) 10.5 (6.1) 10.6 (6.1) 10.0 (6.1) 0.141 11.2 (6.1) 11.5 (6.2) <0.0001 8.7 (5.9) 9.0 (5.7) 9.1 (5.9) <0.0001PGA, M(SD) 72.0 (20.0) 71.6 (20.1) 73.6 (19.6) <0.0001 70.5 (20.0) 74.2 (19.3) <0.0001 68.5 (21.0) 72.5 (18.8) 79.1 (18.5) <0.0001ESR, M(SD) 43.2 (28.4) 47.0 (29.1) 35.9 (25.1) <0.0001 45.2 (12.1) 46.5 (28.6) <0.0001 36.0 (26.9) 37.8 (27.6) 38.8 (29.0) <0.0001HAQ, M(SD) 1.9 (0.6) 1.9 (0.6) 2.1 (0.6) <0.0001 1.9 (0.6) 2.1 (0.6) <0.0001 1.5 (0.7) 1.9 (0.6) 2.2 (0.5) <0.0001RF+, N(%) 11,275 (64.3) 8,985 (64.1) 2,113 (62.1) 0.002 3,465 (64.4) 4,267 (65.2) 0.324 1,484 (61.5) 1,669 (62.5) 255 (63.9) 0.58

Good 1-year treatment response, N(%)* 5,271 (34.3) 4,293 (35.5) 917 (30.0) <0.0001 1,665 (34.7) 1,759 (30.1) <0.0001 997 (44.7) 856 (34.9) 97 (26.4) <0.0001*data available for 15,386 participants.

Table 2. Association between baseline depression and treatment response at 12-months.

History of depression SF36 (nMH≤40) EQ5D

OR 95%CI p OR 95%CI p OR 95%CI pUnadjuste

d Depression symptoms (vs. none) 0.75 0.66, 0.85 <0.0001 0.79 0.69, 0.92 0.002 - - -

Moderate depression symptoms (vs. none) - - - - - - 0.76 0.63, 0.91 0.003Extreme depression symptoms (vs. none) - - - - - - 0.51 0.38, 0.69 <0.0001

Adjusted* Depression symptoms (vs. no depression) 0.80 0.69, 0.92 0.002 0.90 0.77, 1.18 0.177 - - -Moderate depression symptoms (vs. none) - - - - - - 0.85 0.69, 1.04 0.105Extreme depression symptoms (vs. none) - - - - - - 0.62 0.45, 0.87 0.005

* adjusted for age, gender, disease duration, baseline DAS28, number of comorbidities, baseline HAQ. SF36 Medical Outcomes Survey 36-item Short Form. nMH normed Mental Health subscale. OR odds ratio. CI confidence interval. N Number of participants included in analysis.

10


History of depression

In comparison to patients without a history of depression, logistic regression

indicated that patients reporting a history of depression have reduced odds

(OR=0.80, 95%CI: 0.69-0.92) of having a good treatment response by 1-year

follow-up after adjusting for covariates (table 2).

Using multilevel longitudinal models, patients reporting a history of depression

reported significantly lower levels of baseline DAS28 (B=-0.07, 95%CI:-0.12, -0.02)

but a significantly lower rate of improvement in DAS28 over time in comparison to

patients without a history of depression (table 3). Those without a history of

depression reported a total improvement in DAS28 of -0.4 at 1-year, whereas.

patients with a history of depression reported a decrease in DAS28 score of -0.36

between baseline and 1-year follow-up (table 3). This significant interaction effect is

displayed graphically in figure 2.

Supplementary tables t3-t6 show the results of the multilevel longitudinal analyses

examining the relationship between history of depression status and TJC, SJC,

PGA and ESR outcomes respectively. Patients without a history of depression show

significantly reduced improvement in all components over time in comparison to

patients with a history of depression.

SF36 nMH subscale

In comparison to patients scoring ≤40 on the SF36 nMH subscale, logistic

regression analysis revealed that those scoring >40, had no significant difference in

the odds of having a good treatment response at one-year follow-up (table 2).

According to multilevel longitudinal analysis, there were no differences in baseline

DAS28 levels between those scoring ≤40 and >40 on the nMH subscale, although

patients scoring ≤40 reported a significantly reduced rate of improvement in DAS28

over time in comparison to those scoring >40. Whereas patients scoring >40 on the

nMH reduce in DAS28 scores by -0.42 at 1-year, patients scoring ≤40 show an

overall improvement in DAS28 of -0.40 by one-year follow-up (table 3). This

significant interaction effect is shown graphically in figure 2.

Supplementary tables t3-t6 show the results of the multilevel analyses examining

the relationship between nMH status and TJC, SJC, PGA and ESR outcomes

11


respectively. Depressive symptomatology according to the SF36 nMH subscale was

not associated with change in PGA or ESR scores over time, however patients

scoring ≤40 showed reduced improvements in TJC and SJC outcomes in

comparison to patients scoring >40.

EQ5D

Logistic regression analysis adjusting for covariates, reveals no significant

difference in odds of having a good treatment response between patients reporting

no depression symptoms and those reported moderate symptoms (OR=0.85,

95%CI: 0.69-1.04). In comparison to patients reporting no depression symptoms,

those reporting extreme depression symptoms had a significantly reduced odds of a

good treatment response at 1-year follow-up (OR=0.62, 95%CI: 0.45-0.87) (table

2).

Results of longitudinal multilevel analyses reveal no significant difference between

depression symptom groups and baseline levels of DAS28, however in comparison

to patients with no depression symptoms at baseline, those with moderate and

extreme symptoms show significantly reduced rate of improvement over time. In

comparison to patients with no symptoms of depression according to the EQ5D,

who improve by -0.38 at 1-year follow-up, patients with some symptoms and

extreme symptoms report reductions in DAS28 of -0.34 and -0.32 respectively at

one-year follow-up (table 3). The significant interaction between depression

symptoms and follow-up timepoint is displayed graphically in figure 2.

Supplementary tables t3-t6 show the results of the multilevel analyses examining the

relationship between EQ5D status and TJC, SJC, PGA and ESR outcomes

respectively. In comparison to patients with no symptoms of depression at baseline,

those with moderate symptoms show significantly reduced improvements in TJC,

SJC, PGA and ESR over time. In comparison to patients with no symptoms of

depression at baseline, those with extreme symptoms show significantly reduced

improvements in TJC, SJC and ESR over time.

12


Table 3.Association between baseline depression, and DAS28 outcomes over 12-month follow-up.

History of depression SF36 (nMH ≤40) EQ5DB 95%CI p B 95%CI p B 95%CI p

UnadjustedDepression symptoms (vs. none) 0.03 -0.02, 0.08 0.294 0.14 0.09, 0.19 <0.0001 - - -Moderate depression symptoms (vs. none) - - - - - - 0.25 0.18, 0.32 <0.0001Extreme depression symptoms (vs. none) - - - - - - 0.50 0.36, 0.64 <0.0001Follow-up number -0.20 -0.21, -0.20 <0.0001 -0.21 -0.21, -0.20 <0.0001 -0.19 -0.20, -0.18 <0.0001Depression*follow-up 0.02 0.02, 0.03 <0.0001 0.01 0.00, 0.01 0.003 - - -Moderate depression (vs. none) - - - - - - 0.02 0.01, 0.03 <0.0001Extreme depression (vs. none) - - - - - - 0.03 0.01, 0.05 <0.0001Adjusted^Depression symptoms (vs. none) -0.07 -0.12, -0.02 0.011 0.03 -0.02, 0.08 0.276 - - -Moderate depression symptoms (vs. none) - - - - - - 0.04 -0.04, 0.11 0.364Extreme depression symptoms (vs. none) - - - - - - 0.11 -0.04, 0.26 0.144Follow-up number -0.20 -0.21, -0.20 <0.0001 -0.21 -0.21, -0.20 <0.0001 -0.19 -0.20, -0.19 <0.0001Depression symptoms*follow-up 0.02 0.02, 0.03 <0.0001 0.01 0.00, 0.01 0.002Moderate depression symptoms (vs. none) - - - - - - 0.02 0.01, 0.03 <0.0001Extreme depression symptoms (vs. none) - - - - - - 0.03 0.01, 0.05 0.001^model adjusted for age, gender, disease duration, comorbidities and baseline HAQ. B unstandardized coefficient.SF36 Medical Outcomes Survey 36-item Short Form. nMH normed Mental Health subscale.

13


Figure 2. Graphical representation of fully-adjusted interactions between baseline depression symptoms and time on DAS28 outcomes over 12-month follow-up.

14


DISCUSSIONThis study found symptoms of depression at baseline to be associated with reduced

long-term odds of reaching clinical remission in patients receiving their first biologic

drug. This supports previous evidence from US and Norwegian demonstrating

reduced likelihood of reaching remission in patients with symptoms of depression at

treatment initiation [9,11]. We also identified prospective associations between

baseline depression symptom status and disease activity, with depression

symptoms contributing to increased DAS28 over the 12-month follow-up, and

impacting change in DAS28 in response to treatment. Examination of the DAS28

components identified associations between depression and both subjective and

objective aspects of disease activity; effect sizes did not differ between subjective

and objective outcomes, contradicting previous research findings emphasising the

relationship between depressive symptoms and subjective experiences of disease

[3,11,23].

There are several explanations for this novel finding. Firstly, depression is known to

impact health behaviours such as medication adherence [24], and non-adherence

to biologics has been shown to reduce DAS28 treatment response [25]. Whilst

adherence data is not collected for all contributors to the BSRBR-RA databset and

not available for inclusion in this paper, the role of adherence as a mechanism for

this relationship is a valuable area for future research. Secondly, there may be a

biological explanation for these findings. Systemic inflammation and elevated

cytokines typically associated with RA disease manifestation and disease severity

are also identified in people with depressive disorder [26–28]. Finally, the large

sample size available for this analysis may have provided sufficient statistical power

to identify small effect sizes typically unobservable in smaller datasets.

We identified differential effects of symptoms of depression symptoms on

rheumatological outcomes, based on the depression assessment method. Whereas

a history of depression and EQ5D categories were largely predictive of all assessed

outcomes, either showing a main effect or modifying change over time, the SF36

was not associated with ESR or PGA. This may be due to these assessments

representing different elements of mental health. Ticking a depression comorbidity

tick box may indicate a lifetime history of depression, or exposure to mental health

15


treatment, however it provides no timeframe or qualifications for endorsement [29].

As the history of depression assessment may include people who have previously

received treatment for depression, they may not be experiencing current

symptomatology. This measure should be viewed as lifetime depression

prevalence, rather than presence of current symptomatology.

The SF36, alternatively, contains multiple items covering a range of psychological

symptoms, including happiness, nervousness, calmness, tiredness and participation

in social activities [5] and is framed to detect a change from normality within the last

month. It may represent a more nuanced perspective of mental health, including

positive and negative affect, as well as psychosomatic and behavioural symptoms

often associated with chronic illness. We used thresholds based on a validation

study [30], but the high prevalence of “depression” measured on the SF-36

suggests a lack of specificity which may have reduced effect sizes due to

measurement error [31]. The EQ5D assesses current depressive symptomatology,

and although by no means a diagnostic test for depression, representing moderate

sensitivity and specificity, the low proportion of patients reporting “extreme

depression/anxiety” is lower than typical prevalence estimates of depression in RA

[1].

This study has used appropriate longitudinal data analysis methodology to examine

the long-term relationship between symptoms of depression and biologic treatment

response. There is a shortage of high-quality longitudinal investigation in this field,

and the evidence that does exist is limited to studies with highly selected samples,

suboptimal depression assessments, inadequate adjustment for confounding

variables, and inappropriate analysis methodologies [3]. The current study uses the

largest prospective observational biologics registry in the world to examine the

impact of depression symptoms on outcomes in real-world patients undergoing

biologic treatment. Our results are therefore externally valid, representing patients

prescribed biologics across the UK; a diverse population.

There are limitations to consider when interpreting these findings. Although

providing several interpretations of depression, none of the measurement tools

available for baseline depression are “gold-standard” indicators of the presence of

diagnostically ascertained depression. Due to the scarcity with which validated

16


screening tools or diagnostic interviews are utilised to measure depression in RA

research [4], the opportunity to compare three methods in the current paper is

helpful, however given the high prevalence and impact of depression on disease

outcomes, symptoms of depression should be routinely measured in

rheumatological practice.

We did not adjust our models for treatment type, or previous failure with

conventional DMARDs. As all patients are receiving biologics and there is no well-

established association between different types of biologic or DMARD on our

dependent or independent variables, we chose not to include treatment type as a

confounder in our models. No data were available on concurrent mental health

treatment, and it is likely that some patients may have been receiving therapy or

antidepressant treatments which may reduce our observed effects.

These results contribute to the growing body of literature highlighting the role

depression plays in predicting long-term health outcomes and treatment response in

RA. These findings have several implications. Repeated screening and

management of mental disorder should be undertaken as part of clinical care.

Biologics are expensive [32], and poor treatment response can result in switching

biologics, which can result in further costs [33]. Depression should therefore be

routinely measured in RA clinical trials, and in clinical practice.

In conclusion, experiencing symptoms of depression at the start of biologics

treatment is associated with reduced treatment response, impacting change over

time in disease activity. The management of symptoms of depression in routine

care is NICE recommended [34], and depression is treatable within the context of

long-term physical health conditions [35–37]. Future research examining the impact

of mental health intervention for physical health outcomes may identify whether

effectively managing depression can improve treatment response in RA.

Key Messages

1. Depression at baseline contributes to approximately 30% reduced odds of

good biologics treatment response.

2. Depression is associated with reduced change in DAS28 over time in

response to biologics.

17


Financial Acknowledgement

This paper represents independent research part-funded by the National Institute for

Health Research (NIHR) Biomedical Research Centre at South London and

Maudsley NHS Foundation Trust and King’s College London. The views expressed

are those of the authors and not necessarily those of the NHS, the NIHR, the

Department of Health or the British Society for Rheumatology.

The BSRBR-RA is a UK-wide national project to investigate the safety of biologic

agents in routine medical practice. This work was supported by the British Society for

Rheumatology (BSR), which receives restricted income from UK pharmaceutical

companies, presently Abbvie, Celltrion, Hospira, Pfizer, UCB, Samsung and Roche,

and in the past Swedish Orphan Biovitrum and MSD. All decisions concerning

analyses, interpretation and publication are made autonomously of any industrial

contribution.

Conflict of Interest

The authors disclose no conflict of interest

REFERENCES

[1] Matcham, F., Rayner, L., Steer, S. and Hotopf, M. (2013) The prevalence of

depression in rheumatoid arthritis: a systematic review and meta-analysis.

Rheumatology (Oxford, England), 52, 2136–48.

https://doi.org/10.1093/rheumatology/ket169

[2] Matcham, F., Norton, S., Scott, D.L., Steer, S. and Hotopf, M. (2016)

Symptoms of depression and anxiety predict treatment response and long-

term physical health outcomes in rheumatoid arthritis: secondary analysis of a

randomized controlled trial. Rheumatology (Oxford, England), 55, 268–78.

18


https://doi.org/10.1093/rheumatology/kev306

[3] Rathbun, A.M., Reed, G.W. and Harrold, L.R. (2013) The temporal relationship

between depression and rheumatoid arthritis disease activity, treatment

persistence and response: A systematic review. Rheumatology (Oxford,

England), 52, 1785–94. https://doi.org/10.1093/rheumatology/kes356

[4] Kalyoncu, U., Dougados, M., Daurès, J.-P. and Gossec, L. (2009) Reporting of

patient-reported outcomes in recent trials in rheumatoid arthritis: a systematic

literature review. Annals of the Rheumatic Diseases, 68, 183–90.

https://doi.org/10.1136/ard.2007.084848

[5] Ware, J.E.J. and Sherbourne, C.D. (1992) The MOS 36-Item Short-Form

Health Survey (SF-36): I. Conceptual Framework and Item Selection. Medical

Care, 30, 473–83.

[6] Hurst, N.P., Kind, P., Ruta, D., Hunter, M. and Stubbings, A. (1997) Measuring

health-related quality of life in rheumatoid arthritis: validity, responsiveness and

reliability of EuroQol (EQ-5D). Rheumatology, 36, 551–9.

https://doi.org/10.1093/rheumatology/36.5.551

[7] Callhoff, J., Weiss, A., Zink, A. and Listing, J. (2013) Impact of biologic therapy

on functional status in patients with rheumatoid arthritis--a meta-analysis.

Rheumatology (Oxford, England), 52, 2127–35.

https://doi.org/10.1093/rheumatology/ket266

[8] Nam, J.L., Ramiro, S., Gaujoux-Viala, C., Takase, K., Leon-Garcia, M., Emery,

P. et al. (2014) Efficacy of biological disease-modifying antirheumatic drugs: a

systematic literature review informing the 2013 update of the EULAR

recommendations for the management of rheumatoid arthritis. Annals of the

19


Rheumatic Diseases, 73, 516–28. https://doi.org/10.1136/annrheumdis-2013-

204577

[9] Rathbun, A.M., Harrold, L.R. and Reed, G.W. (2016) A prospective evaluation

of the effects of prevalent depressive symptoms on disease activity in

rheumatoid arthritis patients treated with biologic response modifiers. Clinical

Therapeutics, 38, 1759–72.e3. https://doi.org/10.1016/j.clinthera.2016.06.007

[10] Mitchell, A.J. and Coyne, J.C. (2007) Do ultra-short screening instruments

accurately detect depression in primary care?. A pooled analysis and meta-

analysis of 22 studies. British Journal of General Practice, 57, 144–51.

[11] Michelsen, B., Kristianslund, K., Sexton, J., Hammer, H.B., Fagerli, K.M., Lie,

E. et al. (2017) Do depression and anxiety reduce the likelihood of remission in

rheumatoid arthritis and psoriatic arthritis? Data from the prospective

multicentre NOR- DMARD study. Ann Rheum Dis, 0, 1–5.

https://doi.org/10.1136/annrheumdis-2017-211284

[12] Overman, C.L., Bossema, E.R., van Middendorp, H., Wijngaards-de Meij, L.,

Verstappen, S.M.M., Bulder, M. et al. (2012) The prospective association

between psychological distress and disease activity in rheumatoid arthritis: a

multilevel regression analysis. Annals of the Rheumatic Diseases, 71, 192–7.

https://doi.org/10.1136/annrheumdis-2011-200338

[13] Rathbun, A.M., Harrold, L.R. and Reed, G.W. (2015) Temporal effect of

depressive symptoms on the longitudinal evolution of rheumatoid arthritis

disease activity. Arthritis Care & Research, 67, 765–75.

https://doi.org/10.1002/acr.22515

[14] Ferreira, R.J.O., Duarte, C., Ndosi, M., de Wit, M., Gossec, L. and da Silva,

20


J.A.P. (2017) Suppressing inflammation in rheumatoid arthritis: Does patient

global assessment blur the target? A practice-based call for a paradigm

change. Arthritis Care & Research,. https://doi.org/10.1002/acr.23284

[15] Hyrich, K.L., Watson, K.D., Lunt, M., Symmons, D.P.M. and British Society for

Rheumatology Biologics Register (BSRBR). (2011) Changes in disease

characteristics and response rates among patients in the United Kingdom

starting anti-tumour necrosis factor therapy for rheumatoid arthritis between

2001 and 2008. Rheumatology (Oxford, England), 50, 117–23.

https://doi.org/10.1093/rheumatology/keq209

[16] Deighton, C., Hyrich, K., Ding, T., Ledingham, J., Lunt, M., Luqmani, R. et al.

(2010) BSR and BHPR rheumatoid arthritis guidelines on eligibility criteria for

the first biological therapy. Rheumatology, 49, 1197–9.

https://doi.org/10.1093/rheumatology/keq006b

[17] Matcham, F., Norton, S., Steer, S. and Hotopf, M. (2016) Usefulness of the

SF-36 Health Survey in screening for depressive and anxiety disorders in

rheumatoid arthritis. BMC Musculoskeletal Disorders, 17, 224.

https://doi.org/10.1186/s12891-016-1083-y

[18] National Institute for Health and Care Excellence. (2009) Depression in adults

with a chronic physical health problem: recognition and management (clinical

guideline CG91) [Internet]. NICE, http://www.nice.org.uk/guidance/cg91.

[19] Fransen, J. and van Riel, P.L.C.M. (2005) The Disease Activity Score and the

EULAR response criteria. Clinical and Experimental Rheumatology, 23, S93–

9.

[20] Kwok O, Underhill, A., Berry, J., Luo, W., Elliott, T. and Yoon, M. (2008)

21


Analyzing longitudinal data with multilevel models: An example with individuals

living with lower extremity intra-articular fractures. Rehabilitation Psychology,

53, 370–86. https://doi.org/10.1037/a0012765

[21] Twisk, J.W. (2013) Applied longitudinal data analysis for epidemiology: a

practical guide. Cambridge University Press.

[22] Bruce, B. and Fries, J.F. (2005) The Health Assessment Questionnaire (HAQ).

Clinical and Experimental Rheumatology, 23, 14–8.

[23] Matcham, F., Ali, S., Irving, K., Hotopf, M. and Chalder, T. (2016) Are

depression and anxiety associated with disease activity in rheumatoid arthritis?

A prospective study. BMC Musculoskeletal Disorders, 17.

https://doi.org/10.1186/s12891-016-1011-1

[24] DiMatteo, M.R., Lepper, H.S. and Croghan, T.W. (2000) Depression is a risk

factor for noncompliance with medical treatment: meta-analysis of the effects

of anxiety and depression on patient adherence. Archives of Internal Medicine,

160, 2101–7. https://doi.org/ioi90679 [pii]

[25] Bluett, J., Morgan, C., Thurston, L., Plant, D., Hyrich, K.L., Morgan, A.W. et al.

(2015) Impact of inadequate adherence on response to subcutaneously

administered anti-tumour necrosis factor drugs: results from the Biologics in

Rheumatoid Arthritis Genetics and Genomics Study Syndicate cohort.

Rheumatology, Oxford University Press. 54, 494–9.

https://doi.org/10.1093/rheumatology/keu358

[26] Liu, Y., Ho, R.C.-M. and Mak, A. (2012) Interleukin (IL)-6, tumour necrosis

factor alpha (TNF-α) and soluble interleukin-2 receptors (sIL-2R) are elevated

in patients with major depressive disorder: a meta-analysis and meta-

22


regression. Journal of Affective Disorders, 139, 230–9.

https://doi.org/10.1016/j.jad.2011.08.003

[27] Dowlati, Y., Herrmann, N., Swardfager, W., Liu, H., Sham, L., Reim, E.K. et al.

(2010) A meta-analysis of cytokines in major depression. Biological Psychiatry,

67, 446–57. https://doi.org/10.1016/j.biopsych.2009.09.033

[28] Cutolo, M., Kitas, G.D. and LCM van Riel, P. (2014) Burden of disease in

treated rheumatoid arthritis patients: Going beyond the joint. Seminars in

Arthritis and Rheumatism, 43, 479–88.

https://doi.org/10.1016/j.semarthrit.2013.08.004

[29] Buck, H.G., Akbar, J.A., Zhang, S.J., Bettger, J.A.P., Buck, H.G., Akbar, J.A. et

al. (2013) Measuring comorbidity in cardiovascular research: A systematic

review. Nursing Research and Practice, 2013, 1–11.

https://doi.org/10.1155/2013/563246

[30] Matcham, F., Norton, S., Steer, S. and Hotopf, M. (2016) Usefulness of the

SF-36 Health Survey in screening for depressive and anxiety disorders in

rheumatoid arthritis. BMC Musculoskeletal Disorders, 17.

https://doi.org/10.1186/s12891-016-1083-y

[31] Hotopf, M., Chidgey, J., Addington-Hall, J. and Ly, K.L. (2002) Depression in

advanced disease: a systematic review Part 1. Prevalence and case finding.

Palliative Medicine, 16, 81–97. https://doi.org/10.1191/0269216302pm507oa

[32] The British Society for Rheumatology. (2016) Biologic drugs. [Internet].

http://www.rheumatology.org.uk/patient_information/patient_information/

biologic_drugs.aspx.

23


[33] Sullivan, S.D., Alfonso-Cristancho, R., Carlson, J., Mallya, U. and Ringold, S.

(2013) Economic consequences of sequencing biologics in rheumatoid

arthritis: a systematic review. Journal of Medical Economics, 16, 391–6.

https://doi.org/10.3111/13696998.2013.763812

[34] National Collaborating Centre for Mental Health. (2010) Depression in adults

with a chronic physical health problem: treatment and management [Internet].

https://www.nice.org.uk/guidance/cg91/evidence/full-guideline-243876061.

[35] Matcham, F., Rayner, L., Hutton, J., Monk, A., Steel, C. and Hotopf, M. (2014)

Self-help interventions for symptoms of depression, anxiety and psychological

distress in patients with physical illnesses: A systematic review and meta-

analysis. Clin. Psychol. Rev. p. 141–57.

https://doi.org/10.1016/j.cpr.2014.01.005

[36] Fiest, K.M., Hitchon, C.A., Bernstein, C.N., Peschken, C.A., Walker, J.R.,

Graff, L.A. et al. (2017) Systematic review and meta-analysis of interventions

for depression and anxiety in persons with rheumatoid arthritis. Journal of

Clinical Rheumatology, 1. https://doi.org/10.1097/RHU.0000000000000489

[37] Rayner, L., Price, A., Evans, A., Valsraj, K., Higginson, I.J. and Hotopf, M.

(2010) Antidepressants for depression in physically ill people. Cochrane

Database Syst. Rev. p. CD007503.

https://doi.org/10.1002/14651858.CD007503.pub2.

24

· web viewthe relationship between depression and biologic treatment response in rheumatoid...

Documents